The Cinchona Primary Amine-Catalyzed Asymmetric Epoxidation and

Apr 18, 2013 - Moreover, the use of the pseudoenantiomeric catalyst 1b enabled facile synthesis of enantiomeric product ent-12f with equal efficiency ...
0 downloads 11 Views 3MB Size
Article pubs.acs.org/JACS

The Cinchona Primary Amine-Catalyzed Asymmetric Epoxidation and Hydroperoxidation of α,β-Unsaturated Carbonyl Compounds with Hydrogen Peroxide Olga Lifchits, Manuel Mahlau, Corinna M. Reisinger, Anna Lee, Christophe Farès, Iakov Polyak, Gopinadhanpillai Gopakumar, Walter Thiel, and Benjamin List* Max-Planck-Institut für Kohlenforschung, Kaiser-Wilhelm-Platz 1 D-45470, Mülheim an der Ruhr, Germany S Supporting Information *

ABSTRACT: Using cinchona alkaloid-derived primary amines as catalysts and aqueous hydrogen peroxide as the oxidant, we have developed highly enantioselective Weitz−Scheffer-type epoxidation and hydroperoxidation reactions of α,β-unsaturated carbonyl compounds (up to 99.5:0.5 er). In this article, we present our full studies on this family of reactions, employing acyclic enones, 5−15-membered cyclic enones, and α-branched enals as substrates. In addition to an expanded scope, synthetic applications of the products are presented. We also report detailed mechanistic investigations of the catalytic intermediates, structure−activity relationships of the cinchona amine catalyst, and rationalization of the absolute stereoselectivity by NMR spectroscopic studies and DFT calculations.



INTRODUCTION

The resulting peroxyenamine F then undergoes epoxide ring closure with the expulsion of water, much like in the original Weitz−Scheffer system. This catalytic platform proved to be highly efficient for the epoxidation of unencumbered α-unbranched α,β-unsaturated aldehydes, as demonstrated by the seminal works of Jørgensen and later MacMillan who used various chiral secondary amines as catalysts.5 Our group has also contributed to this reaction in 2008 with the use of asymmetric counteranion-directed catalysis (ACDC), in which an achiral secondary amine catalyst together with the chiral Brønsted acid TRIP were used to induce enantioselectivity.6 Despite the success of secondary amine catalysts in the epoxidation of α-unbranched enals, their extension to more sterically demanding substrates remained a challenge. 7 Furthermore, at the outset of our studies, the enantioselective epoxidation of simple cyclic enones, nonchalcone-type acyclic enones, and α-branched enals presented a general problem, for which neither metal- nor small organic molecule-based catalysis offered highly enantioselective protocols. Our early interest in the use of primary amine catalysts as a tool to tackle difficult substrates prompted us to explore this catalyst class in various transformations.8 For example, in 2006 we developed a highly enantioselective transfer hydrogenation of enones using an α-amino acid ester/chiral phosphate salt.8a Encouraged by this and other seminal studies,9 we sought to apply primary amine catalysis to asymmetric epoxidation of challenging α,β-unsaturated carbonyl compounds. As a guiding principle, we relied on the assumption that the reduced steric requirements of primary amines may overcome the inherent

1

Since the seminal discoveries of Sharpless in the 1980s, catalytic olefin epoxidation has continued to hold a prominent place in modern asymmetric synthesis. Milestone achievements in this field over the last thirty years, including those associated with the names of Juliá and Colonna, Wynberg, Jackson, Jacobsen, Katsuki, Enders, Shi, and Shibasaki, have not only provided reliable synthetic protocols but also identified entirely novel catalytic platforms.2 One of the most fundamental ways to form an oxirane ring is the reaction of basic hydrogen peroxide with electrophilic α,βunsaturated ketones and aldehydes, a process described already in 1921 by Weitz and Scheffer.3 Kinetic studies of this transformation established it to be a two-step addition− elimination process, in which hydrogen (or alkyl) peroxide, deprotonated under basic conditions, adds to the β-carbon of the unsaturated carbonyl compound, forming a β-peroxyenolate intermediate B (Scheme 1a).4 This is followed by an intramolecular ring closure, in which the weak O−O bond is broken and a second C−O bond is formed, expelling a hydroxide or an alkoxide anion as the leaving group. Protonation of the enolate B results in the common sideproduct of the reaction, hydroperoxide D (typically observed in its more stable hemiketal form as shown). Several enantioselective variants of the Weitz−Scheffer epoxidation have been developed to date, which rely on the catalytic activation of the substrate and/or the peroxide reagent. Most recently, it has been recognized that Weitz−Scheffer-type epoxidation is also amenable to covalent aminocatalysis, whereby the carbonyl substrates are activated as iminium ion intermediates E toward conjugate addition by the oxidant (Scheme 1b). © XXXX American Chemical Society

Received: March 5, 2013

A

dx.doi.org/10.1021/ja402058v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society

Article

Scheme 1. Weitz-Scheffer Epoxidation: (a) the Original Mechanism and (b) the Amine-Catalyzed Variant

chalcone-type substrates were broadly suitable, with simple enolizable aliphatic enones still posing a considerable challenge to chemists.2b,d,g,l,14 Inspired by our earlier success with primary amine catalysis,8a we decided to examine trifluoroacetic acid (TFA) salts of the cinchona alkaloid-derived amine 1a in this transformation. Using 3-decen-2-one (2a) as a model substrate and hydrogen peroxide as an inexpensive, environmentally benign, and easily handled oxidant, we observed complete conversion of the starting material to two isolable products: cyclic peroxyhemiketal 3a, a known intermediate and common byproduct in Weitz−Scheffer-type epoxidations,15 and the expected epoxide 4a as the minor product in essentially enantiopure form. Since cyclic peroxides 3 can be quantitatively transformed into the corresponding epoxides 4 under basic conditions,16 and serve as precursors to other useful compounds, we decided to first optimize the hydroperoxidation reaction. The ratio of the intermediate peroxyhemiketal 3 and the terminal epoxide product 4 was found to mainly depend on the pKa of the acid cocatalyst, water content, and temperature of the reaction, with weaker acids, higher water concentration, and lower reaction temperature favoring the hydroperoxidation product and decelerating the formation of the epoxidation product.17 With the optimal conditions in hand, we evaluated the scope of the reaction, employing various aliphatic enones 2 (Table 1). A wide variety of substrates was found to be compatible with the reaction, affording the desired cyclic peroxyhemiketals 3 as the major products, which could be easily separated from the epoxide side-products 4 and isolated with reasonable yields and invariably excellent enantioselectivities. Both linear and branched aliphatic substituents R1 were well tolerated (entries 1−4), although the degree of branching affected the hydroperoxide/epoxide product distribution. Functional groups including an isolated double bond (entry 5), acid-labile protected alcohol moieties (entries 6−7), an ester (entry 8), a pendant ketone (entry 9), and an unprotected hydroxy group (entry 10) were acceptable. Notably, the second ketone group in substrate 2j gave rise to an equilibrium mixture of two cyclic peroxyhemiketal products, containing five- and six-membered dioxolane rings.17 Five-membered peroxyhemiketal 3j′ was found to exist as the typical mixture of two diastereomers, while only one diastereomer of the six-membered isomer 3j″ could be detected, whose relative configuration was putatively assigned. Various groups R2, which render substrates 2 enolizable, all participated in the reaction smoothly (entries 13−17), although higher steric demand of the R2 group was found to decrease reactivity and increasingly favor the formation of the epoxide products 4. Surprisingly, substrate 2m bearing an isobutyl group almost exclusively afforded the epoxide product (entry 14).

difficulties of secondary amines in generating congested covalent iminium ion intermediates (Scheme 2).10 Scheme 2. Challenging Substrate Classes Prior to our Studies and our Proposed Catalytic Platform

Thus, we speculated that if primary amine salts are compatible with oxidizing conditions, they could provide a general solution for the epoxidation of sterically demanding substrates. Indeed, both iminium and enamine catalysis employing chiral primary amines have started to emerge as a powerful strategy in catalytic asymmetric transformations in recent years.2e,10,11 Initial investigations using the salts of cinchona alkaloid-derived primary amine catalyst 1a11a−c,g,12 confirmed our expectations, and we have recently developed and communicated highly enantioselective epoxidation and hydroperoxidation reactions of a variety of cyclic and acyclic enones as well as α-branched enals.8b,c,e Subsequent to our initial report,8c Deng et al. independently published related alkylperoxidations of acyclic enones using catalyst 1a.11e All of the reactions employ catalyst 1a or its close analogues and feature both generality and a remarkable consistency in the stereochemical outcome. Herein, we present our full studies on this family of reactions, including an expansion of the scope, synthetic applications, and mechanistic investigations.



RESULTS AND DISCUSSION Epoxidation and Hydroperoxidation of Acyclic Enones.8b At the outset of our studies in 2008, general and highly enantioselective strategies for the catalytic epoxidation of simple, α,β-unsaturated ketones were scarce.2j,13 Indeed, despite the wealth of epoxidation methodologies, only B

dx.doi.org/10.1021/ja402058v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society

Article

Table 1. Catalytic Asymmetric Hydroperoxidation of Enones

Table 2. Catalytic Asymmetric Epoxidation of Enones

a All reactions performed on 0.5−1 mmol scale. bDetermined by 1H NMR analysis of the crude reaction mixture. cDetermined by GC with a chiral stationary phase after conversion to the corresponding epoxide with 1 N NaOH (1 equiv) in THF. dConducted on 3 mmol scale. e 69% conversion determined by 1H NMR. fer of the initially formed epoxide 4k. g96% conversion determined by 1H NMR.

a All reactions performed on 0.5−1 mmol scale. bDetermined by 1H NMR analysis of the crude reaction mixture. cDetermined by GC with a chiral stationary phase. dUsing 10 mol % [1b·2TFA]. eSum of 3q and the corresponding peroxyhemiketal 4q at 94% conversion determined by 1H NMR. fReduced yield due to the volatility of the product. gNaOEt (1 equiv) was used instead of NaOH. hSecond step was omitted. iSum of 4k and the corresponding peroxyhemiketal 3k at 69% conversion determined by 1H NMR. jUsing 20 mol % [1a·2TFA].

A slight modification of the reaction conditions and an inclusion of a basic workup of the crude reaction mixture with 1N NaOH allowed for an efficient generation of epoxide products 4 (Table 2). Gratifyingly, good to excellent yields and very good enantioselectivities were obtained for a range of substrates of varying steric demands. Importantly, using the pseudoenantiomeric catalyst 1b, the opposite enantiomers of the epoxide products could be accessed with nearly equal efficiency (entries 1−2 and 3−4). Interestingly, both substrates 2r bearing a bulky R1 group (entry 7) and 2m bearing a bulky R2 group (entry 21) afforded the respective epoxide products with essentially perfect enantiocontrol (er >99.5:0.5). In addition, epoxide 4m (entry 21) was formed as the only product of the reaction, whereby no corresponding peroxyhemiketal product could be detected. All of the functional groups which were found to be compatible in the hydroperoxidation procedure (Table 1) were also suitable for epoxidation (entries 10−16). However, the hydrolyzing conditions of aqueous NaOH workup required in the epoxidation protocol resulted in partial saponification of

product 4i (entry 15) which bears an ethyl ester group. This problem could be overcome by using NaOEt in ethanol instead of aqueous NaOH in the workup. In case of the ketone-bearing substrate 2j (entry 16) the basic workup was omitted since it did not improve the yield of epoxide 4j, presumably due to a competing base-catalyzed aldol reaction. Thus, epoxide 4j was isolated from the hydroperoxide−epoxide product mixture in 40% yield and 98.5:1.5 er (entry 16). Low conversion (69%) of 9-hydroxy-3-nonen-2-on 2k containing an unprotected hydroxyl group, which could not be improved by an extended reaction time, resulted in only moderate yield of the corresponding products, although the enantioselectivity remained high (98:2 er, entry 17).18 We also investigated the influence of olefin geometry on the enantioselectivity of the epoxidation reaction by submitting C

dx.doi.org/10.1021/ja402058v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society

Article

Thus, both substituents R1 and R2 are limited to alkyl groups, a phenomenon also observed by Deng and colleagues in the 1acatalyzed hydroperoxidation of α,β-unsaturated ketones with tert-butyl peroxide and related oxidants.11e Similarly, literature analysis confirmed that the aromatic R2 substituents (Scheme 4) are only very rarely tolerated in reactions catalyzed by cinchona alkaloid-derived primary amines.22 In this respect, our catalytic epoxidation complements conventional methods, such as metal-catalyzed or Juliá-Colonna epoxidations. Despite the reaction’s tolerance of remote protected hydroxy functionalities, a benzyloxy group in the γ position proved deleterious (substrate 5f). Neither substitution at the α-position nor extended conjugation of the double bond is tolerated, as illustrated by the reaction of 1-acetylcyclohexene (5g) and βionone (5h), respectively. 4-Oxoenoates 5i−j also did not undergo epoxidation, although their use in related transformations via asymmetric iminium ion catalysis is well precedented in the literature. Surprisingly, terminal enones were unsuitable in the reaction (substrate 5k). We speculate that in the case of more active Michael acceptors, conjugate addition of catalyst 1a to form β-amino ketones is favored over condensation with the ketone group, especially given that hydrogen peroxide (presumably in its neutral form) is a weak nucleophile to compete with. To probe this hypothesis, we conducted the epoxidation of our model substrate 3-decen-2one (2a) in the presence of the terminal enone 5k (1:1 ratio), adjusting the reaction parameters to imitate the optimal conditions. Only epoxide 4a was generated, albeit at a significantly lower rate. This is consistent with the assumption that the catalyst is captured as a β-amino ketone by 5k and only partly released from this “catalyst resting state”, keeping its absolute concentration low.23 Synthetic Utility of the Obtained Epoxides and Hydroperoxides. To demonstrate the utility of our hydroperoxidation and epoxidation reactions, we also explored several synthetic applications of the corresponding products 3 and 4. We recognized that a simple reduction of the O−O bond in hydroperoxides 3 could deliver valuable β-hydroxyketones 6. Importantly, organocatalytic aldol approaches to these compounds are limited by the need to use unbranched enolizable aldehydes, which are prone to self-aldolization and typically lead to lower yields and inferior enantioselectivities. In contrast, hydroperoxidation of enones 2 followed by reduction, which represents a formal enantioselective hydration of the α,βunsaturated ketones, could furnish β-hydroxyketones 6 from simple starting materials and draw upon the high enantioselectivity of our hydroperoxidation procedure. Based on this rationale, we developed a one-pot protocol which consists of asymmetric hydroperoxidation, followed by the addition of triethylphosphite24 to the crude reaction mixture. Various aldol-like products 6 were thus obtained in good yields and excellent enantioselectivities (Table 3). Alternatively, the crude peroxyhemiketals can also be reduced by catalytic hydrogenation with equal efficiency and without any erosion of the enantiomeric excess (Table 3, entry 2).26 The existence of potent natural products which contain optically active five-membered 1,2-dioxolanes27 and the noticeable lack of methodologies for their asymmetric synthesis in the chemical literature28 stimulated us to seek approaches to this substrate class based on our methodology. We envisaged optically active peroxyhemiketals 3 as precursors for the

isomerically pure enones (E)-2b and (Z)-2b to the optimal reaction conditions (Scheme 3). Remarkably, both substrates Scheme 3. Stereoconvergence in the Asymmetric Epoxidation

furnished the same enantiomer of trans-epoxide 4b with excellent enantioselectivity and yield, indicating complete stereoconvergence. This behavior suggests that E/Z mixtures of enones can be conveniently employed as epoxidation substrates. GC-MS analysis of samples taken from the epoxidation of pure (Z)-2b revealed the formation of (E)-2b, indicating rapid isomerization of the (Z)-enone to the corresponding (E)isomer under the reaction conditions. (Z)-2b was also found to isomerize in the presence of catalytic salt [1a·2 TFA] alone, without any assistance of hydrogen peroxide. This suggests that isomerization may take place via a dienamine intermediate, which allows free rotation about the carbon−carbon single bond. We, and others, have proposed an analogous mechanism in earlier studies on the asymmetric transfer hydrogenation of α,β-unsaturated aldehydes with a secondary amine catalyst, where similar stereoconvergence was observed.19,20 Several substrate classes currently represent a limitation of the method and are summarized in Scheme 4. Enones Scheme 4. Current Limitations in Enone Hydroperoxidation and Epoxidation

conjugated with an aromatic group (5a−c, 5e) all gave the corresponding epoxide products in 12) enones. In contrast, no hydroperoxide intermediates could be isolated with enals, cyclopentenones, and cyclohexenones. A combination of enamine and Brønsted acid catalysis is believed to promote the ring closure of B1/B2 to the epoxide intermediate C1/C2.45 Subsequent hydrolysis liberates the epoxide product and regenerates the catalyst. An important difference from the Weitz−Scheffer epoxidation concerns the reversibility of the oxa-Michael addition of hydrogen peroxide (step A→B, Scheme 8). The basic conditions of Weitz−Scheffer epoxidations create a hydroperoxide anion as the active nucleophile, whose addition to unsaturated carbonyls has been confirmed by kinetic data to be fast and reversible.38d,38e Under our reaction conditions in the presence of catalyst salts, which include 2 equiv of a Brønsted acid, neutral hydrogen peroxide rather than the hydroperoxide anion is expected to be the active nucleophile. Consequently, the conjugate addition is most likely irreversible and constitutes the enantiodetermining step of the catalytic cycle.46 This is supported by the high enantiomeric excess of the isolable hydroperoxy intermediates 3 and 13 (er >95:5), which would be expected to be racemic under a reversible mechanistic regime.47 To provide comprehensive support for the proposed mechanism and to elucidate the origin of enantioselectivity, we performed mechanistic studies on both enone and enal substrates, employing various experimental and computational approaches.

Table 9. Catalytic Asymmetric Epoxidation of 2-Substituted Acroleins

a

All reactions performed on 0.25 mmol scale. bIsolated yields after NaBH4 reduction. cDetermined by GC with a chiral stationary phase. d Isolated yield of the aldehyde. eThe decreased yield reflects the high volatility of the product.

Diminished yield for substrate 22c (entry 3) reflects the high volatility of the corresponding product 23c. Applying the pseudoenantiomeric form of the amine catalyst 1b and (S)-21a as the catalytic salt in the epoxidation of 22a provided the corresponding (S)-product ent-23a with only 44% yield and a modest enantioselectivity of 71:29 er. However, complementary approaches, which furnish the (S)-products with high enantiomeric excess, were recently reported by the groups of Hayashi7a and Luo.11d I

dx.doi.org/10.1021/ja402058v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society

Article

enone 2a. After H2O2 addition (Scheme 9), iminium ion A could no longer be detected, suggesting its rapid consumption by hydrogen peroxide, although the signal for the free catalyst 1a remained. Already after 1 h reaction time, two new peaks could be detected at m/z 476 and at m/z 494 which match the masses of the iminium ion B derived from 2,3-epoxy-2decanone (4a) and iminium ion C derived from the corresponding β-hydroperoxyketone, which upon hydrolysis affords peroxyhemiketal 3a. LRMS and HRMS data obtained for catalyst 1a and intermediates A−C are summarized in Table 10. The fact that the free catalyst 1a as well as the intermediate products of hydroperoxidation (B) and intramolecular epoxide closure (C) could be easily detected under the reaction conditions, while the iminium ion A remained below the detection limit, suggests that the initial formation of the iminium intermediate A could be the rate-limiting step of the reaction.17 When analyzing the reaction at 50 °C, we were particularly intrigued by the appearance of additional signals at m/z 340, 492, and 510, which presumably correspond to the oxidized catalyst (species 1a+O) and iminium ions B+O and C+O that incorporate this oxygenated catalyst (Scheme 9b and Table 10). After 12 h of reaction time, these peaks gained in intensity, while the parent peaks 1a, B and C (m/z 324, 476 and 494, respectively) showed considerable loss of intensity, suggesting catalyst oxidation.17 It is noteworthy that a doubly oxidized catalyst species, i.e. a mass incorporating two oxygens (m/z 356) could not be detected. Since all of the reactions with the amine catalyst 1a proceed to completion and the epoxidation products are formed with outstanding enantioselectivities, the observed in situ oxidation of the catalyst raised interesting mechanistic questions. Specifically, we wondered if the oxidized catalyst 1a+O is active, and hence if the iminium ions B+O and C+O are catalytically relevant intermediates. Their formation suggested to us that the primary amine functionality of the catalyst 1a cannot be involved in the oxidation to 1a+O. We therefore hypothesized that species 1a+O most likely arises from the N-oxidation of the most nucleophilic quinuclidine nitrogen. Toward this end, we synthesized 9-amino(9-deoxy)epiquinine N-oxide 24 using mCPBA (Scheme 10) and could unambiguously confirm the site of N-oxidation by 15N-HMBC analysis.17 With an authentic sample of 24 in hand, we next analyzed the catalyst species, which remained at the end of the epoxidation reaction shown in Scheme 9. Although analysis of the crude reaction revealed a complex mixture of compounds containing the quinine scaffold, which presumably correspond to various catalytic intermediates, direct elution on silica gel allowed us to isolate a single species of significantly higher polarity. Characterization of this product confirmed it to be identical to 24. In addition, when we submitted 1a to the standard conditions of epoxidation in the absence of any enone substrate, we observed a clean and quantitative formation of the bis-trifluoroacetic acid salt of 24. Having thus established the identity of 1a+O as the N-oxide 24, we examined its chemical reactivity. Oh and Ryu have recently shown that tertiary amine N-oxides can act as oxygen transfer agents to effect asymmetric epoxidation of enones at 100 °C.50 However, mixing an excess (1.5 equiv) of 24 with enone 2a in the absence of hydrogen peroxide at 50 °C did not give any oxidation products, ruling out the ability of 24 to act as an oxidant under our standard reaction conditions.

Scheme 8. Proposed Catalytic Cycles for the Epoxidation of Enone and Enal Substrates with Catalysts 1

Structure−Activity Relationship of Modified Catalyst 1a. We first verified that the presence of the primary amine moiety in 1a is indeed crucial for the observed reaction rate and enantioselectivity, supporting a mechanism proceeding through iminium activation and not via other activation modes, such as general base catalysis or phase-transfer catalysis.48 This is further corroborated by the complementarity of scope observed in reactions catalyzed by 1a and in phase transfer catalysis. While steric congestion around the carbonyl group (e.g., αsubstituents in the cyclohexenone series) is detrimental in our reactions, an opposite effect is observed under phase-transfer catalysis in comparable systems.49 ESI-MS Studies and Catalyst N-Oxidation. To provide further support for covalent catalytic intermediates and to follow the fate of the amine catalyst 1a, we performed electrospray ionization mass spectroscopy (ESI-MS) studies using the epoxidation of acyclic enone 2a as a model system (Scheme 9). Samples of the crude reaction mixture taken at various time points over 24 h were analyzed, and the selected spectra are shown in Scheme 18 in the Supporting Information.17 The ESI(+)-MS spectrum of a sample taken prior to the addition of hydrogen peroxide (Scheme 9a) revealed the characteristic signals corresponding to the amine catalyst 1a at m/z 324 ([1a+H]+) as well as iminium ion A at m/z 460, formed by the condensation of the amine 1a and J

dx.doi.org/10.1021/ja402058v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society

Article

Scheme 9. ESI(+)-MS Studies on the Epoxidation and Hydroperoxidation of 2a: (a) Spectrum of the Reaction Sample Recorded (a) Prior to H2O2 Addition and (b) at 50 °C after 1 h

Table 10. Low- and High-Resolution Mass Spectrometric Analysis of Catalytic Species in the Epoxidation of Enone 2a

a

HRMS of purified 24.

amine 24 is catalytically active. Furthermore, when using amine 24, epoxides 20a, 23a, and 4a were formed with the same sense of stereoinduction and similar enantio- and diastereoselectivity as with catalyst 1a. These findings suggest that the in situ Noxidation of catalyst 1a creates a small amount of a different, but similarly competent catalytic species, which might catalyze the reaction in parallel with 1a, explaining the high

We next investigated the competence of 24 as a catalyst, submitting it to the epoxidation reactions of enals 19a, 22a, and enone 2a under the corresponding optimized conditions for each substrate, along with a control reaction using amine 1a (Table 11). After 24 h, all substrates underwent epoxidation with a comparable conversion using either catalyst, suggesting that the K

dx.doi.org/10.1021/ja402058v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society

Article

conditions, we could also prepare stable solutions of iminium salts 26 containing 0.5−2 equivalents of Brønsted acids TFA and (R)-TRIP ((R)-21a).

Scheme 10. Synthesis of an Authentic Sample of 9-Amino(9deoxy)epiquinine N-Oxide 24

Scheme 11. Synthesis of Imine 25 and Iminium Salts 26

Table 11. Catalytic Activity of N-Oxide 24

entry

amine catalyst

acid cocatalyst

product

yield [%]a,c

drb

erb,d

1 2 3 4 5 6

1a 24 1a 24 1a 24

(R)-21b (R)-21b (R)-21a (R)-21a TFA TFA

20a 20a 23a 23a 4a 4a

94 99.5 97 90 99 83

93:7 86:14 − − >99:1 >99:1

99:1 (R,S) 97.5:2.5 (R,S) 99:1 (R) 94:6 (R) 98.5:1.5 (S,R) 96.5:3.5 (S,R)

Unfortunately, extensive attempts to crystallize 25 or its various salts did not meet with success. However, a combination of solution-state NMR and DFT calculations allowed for a detailed characterization and conformational assignment of imine 25 and its salts 26.17 Figure 2a represents

a

Determined by 1H NMR analysis. bDetermined by GC with a chiral stationary phase. cRefers to the GC yield of the product 20a, 23a, or 4a. dRefers to the er of the epoxide product 20a (major diastereomer), 23a or 4a (after 1 N NaOH workup). See Tables 2, 8, 12 for precise reaction conditions. TFA = trifluoroacetic acid.

enantioselectivity maintained throughout our epoxidation reactions and complete conversion of the starting materials. Origin of Enantioselectivity. Despite the myriad of synthetic transformations employing catalyst 1a and its related cinchona-based congeners reported in recent years, only a single computational investigation has attempted to rationalize the origin of enantioselectivity imparted by this versatile catalytic platform.51 A number of methodological reports have speculated on the role of the quinuclidine moiety in catalyst 1a suggesting that it acts as a free base in the transition state (TS), coordinating the weakly acidic hydrogen peroxide.5b,c,8e,45 Yet our findings of the excellent catalytic performance by the Noxidized catalyst 24 in which the quinuclidine nitrogen is replaced with a weakly basic N-oxide moiety52 raised intriguing mechanistic questions. We decided to investigate the mechanism of facial selectivity by 1a using a combination of conformational analysis of catalytic intermediates and computational techniques. As outlined earlier, experimental evidence suggested that the conjugate addition of hydrogen peroxide constitutes the enantiodetermining step of our epoxidation reactions. We therefore chose to investigate the chiral iminium adducts A, which are formed by the reversible condensation of catalyst 1a with the carbonyl substrate, and act as acceptors in the first C− O bond-forming conjugate addition (Scheme 8). Toward this end, we found that α-branched enals 19 provide a useful model system, undergoing smooth condensation with amine 1a in the presence of 4 Å molecular sieves. Using enal 19e in a slight (5%) excess to ensure complete consumption of 1a, the corresponding imine 25 could be isolated without purification as a bench-stable amorphous solid containing traces of the remaining 19e (Scheme 11). Employing strictly anhydrous

Figure 2. Structure of the monoprotonated iminium 26 elucidated by NMR techniques (a) and gas-phase density functional calculations (B3LYP/SVP) (b).

the structural features of the monoprotonated iminium 26 obtained by NMR analysis, which was found to be in excellent agreement with an energy-minimized structure derived by DFT calculations (Figure 2b). Similarly to the parent cinchona alkaloids,53 imine 25 was found to have a strong conformational preference in solution, which was further rigidified by protonation. Analysis of salts 26a−e revealed that the presence of different counteranions in different stoichiometries (0.5−2 equiv) did not influence the overall conformation of the iminium cation, although some L

dx.doi.org/10.1021/ja402058v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society

Article

obtained with either chiral or achiral acid cocatalyst in methanol (entries 1−2). These findings suggest that while an acid additive can catalyze the reaction even in a disruptive protic solvent by protonating the imine intermediate of type 25, it has no influence on the stereoselectivity of the reaction if the ions are solvent separated.55 In a parallel line of investigations, we performed the epoxidation of enal 19a in a nondisruptive aprotic solvent but employed Brønsted acid cocatalysts whose conjugate bases form anions of different coordinating strengths. Figure 3 shows

smaller conformational adjustments are possible. Key structural features of 26 include an s-trans conformation adopted by the unsaturated imine moiety, with both double bonds possessing (E)-geometry. Analysis of 15N chemical shifts and 1JH−N coupling constants in monoprotonated salts 26a and 26c revealed full protonation of the quinuclidine nitrogen and partial protonation of the imine nitrogen, consistent with an internal hydrogen bond forming a pseudofive-membered ring N1−C8−C9−N21−H (Figure 2). In the presence of 2 equiv of a Brønsted acid (salts 26b and 26d), this hydrogen bond was found to be undisturbed, with the second site of protonation being the quinoline nitrogen.17 The effect of this internal hydrogen bonding manifests itself in the hindrance of C9−N21 bond rotation and C8−C9 bond rotation (with H8 and H9 adopting an anti relationship). Indeed, the only dynamic moiety was found to be the quinoline group, which undergoes rapid conformational exchange (rotation around the C9−C4′ bond) at room temperature in unprotonated imine 25 and a somewhat hindered rotation in the protonated iminium species 26 as witnessed by corresponding peak broadening and NOESY analysis.17 Although diffusion ordered spectroscopy (DOSY) indicated a tight association of the ammonium cation and the counteranion in salts 26 in aprotic or weakly protic media,17 heteronuclear NOE experiments failed to pinpoint the position of the counteranion in solution. In both TFA and (R)-TRIP salts, a large number of weak cross-peaks from the 19F or 31P nucleus to 1H signals from all three subcomponents of 25 (quinuclidine, quinoline, and aza-diene) with similar intensities was observed, suggesting some flexibility in the ion pairing.54 Nevertheless, the strong influence of the Brønsted acid cocatalyst on the enantioselectivity of our epoxidation reactions suggested involvement of the counteranion in the asymmetric induction. In order to probe this hypothesis further, we examined the effect of ion separation by a protic solvent. To this end, we carried out the epoxidation of enal 19a in methanol, which was confirmed by DOSY analysis to completely separate the ammonium cation and the counteranion in salt 26c without affecting the overall conformation of the protonated imine.17 As demonstrated in Table 12, carrying out the epoxidation in methanol resulted in complete erosion of the reaction’s

Figure 3. The effect of the counteranion coordination strength in catalyst salts [1a·2HX] on the enantioselectivity of enal epoxidation. Conditions A: THF, 50% H2O2 (5 equiv) and B: 1,4-dioxane, 50% H2O2 (1.5 equiv).

the enantiomeric excess of the major trans-diastereomer 20a obtained with each of the counteranions X− in the catalytic salts [1a·2HX] (see Supporting Information for conversion and dr). Approximate relative coordinating strength of the screened counteranions is provided along the x-axis.56 A clear trend can be observed with respect to counteranion association in the catalytic salt [1a·2HX] and the enantioselectivity of the reaction. In particular, the most tightly coordinating anions gave the highest enantiomeric excess, even without possessing chirality themselves. Computational Studies and the Proposed Model of Stereoselectivity. With the structure of the ammonium cation established by physical methods and evidence for the involvement of the counteranion in the asymmetric induction, we turned to computation to complete the model of stereoselectivity. Toward this end, we analyzed the geometries of TSs for the enantiodetermining step, i.e., hydrogen peroxide attack on the iminium ion. We employed DFT throughout. Geometry optimizations were performed using the BP86 functional and the def2-SVP basis set. Single-point calculations at the BP86/def2-SVP optimized geometries were carried out with larger basis sets (def2-TZVP) and different functionals (B3LYP, M06, M06-2X) and with inclusion of continuum solvation and empirical dispersion corrections to arrive at more accurate energetics. Thermal and entropic corrections were evaluated at the BP86/def2-SVP level to convert the calculated energies into free energies. The Supporting Information contains detailed information on the applied computational methods and the numerical results as well as more detailed discussions of the results obtained for the various pathways considered. In the following, we focus on the two major

Table 12. The Effect of an Ion-Separating Solvent on the Epoxidation of 31a with Catalytic Ammonium Salts of 1a

entry 1 2 3 4

acid (R)-21a TFA (R)-21a TFA

solvent MeOH MeOH THF THF

conv. [%]a c

37 78 81 73

dra

er (trans)b

51:49 51:49 87:13 65:35

43:57 46:54 99:1 75:25

a

Determined by GC. bDetermined by GC with a chiral stationary phase. cThe catalyst salt was poorly soluble.

diastereoselectivity and a greatly reduced and reversed enantioselectivity (compare entries 1−2 to entries 3−4), although the conversion remained high (the lower conversion obtained with the salt [1a·(R)-21a] in methanol reflects poor catalyst solubility). More importantly, very similar results were M

dx.doi.org/10.1021/ja402058v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society

Article

Figure 4. Energetically lowest-lying reaction pathways and TS leading to the favored enantiomer (pathway 1, left structure) and the unfavored enantiomer (pathway 2, right structure) computed at the BP86/def2-SVP level. See Table 3, Supporting Information, for further information.

NMR structure of 26 (Figure 2). This coordination also implies that the quinuclidine nitrogen is not directly involved in directing the peroxide attack. There are also considerable differences between TS-1 and TS-2, which are helpful in rationalizing the observed enantioselectivity. In TS-1 the configuration around the newly formed C−O bond is almost perfectly staggered, the anion being roughly oriented toward the positively charged iminium ion. The distance between the closer TFA oxygen and the iminium proton is 3.95 Å. On the other hand, for TS-2, in order to avoid a steric clash with the quinoline moiety, the TFA-peroxide adduct prefers a different orientation. Here, the configuration around the C−O bond is close to eclipsed, and the cation anion distance is increased to 4.13 Å. As the anion is much better accommodated on one side of the iminium ion and directs hydrogen peroxide to the same side, it effectively acts as a mediator, communicating the steric information around the catalysts stereocenter to the rather distant point of attack. Our model predicts that the quinoline moiety should influence enantioselection, but a degree of conformational flexibility is allowed, as the two most stable rotamers around the C9−C4′ bond both favor the same enantiomer. As described previously, such flexibility of the quinoline ring in imine salts 26 was indeed observed by NMR experiments. Furthermore, experiments using the truncated catalyst, which lacks a quinoline moiety, revealed that this group is indeed essential for the stereoinduction.17 Based on our model of stereoselectivity, we propose that the excellent catalytic efficiency of the N-oxidized amine 24 stems from its ability to form structurally similar iminium salts 27, which participate in the enantiodetermining C−O bondforming event (Figure 5). In particular, our data suggest that the N-oxide moiety simply expands the internal hydrogenbonded assembly observed in salts 26 by one oxygen atom, forming a pseudosix-membered ring and thus causing little perturbation in the TS. A similar type of intramolecular hydrogen bonding has been reported in chiral N,N-dioxide amide catalysts.59 In order to generalize our enantioselectivity model to the epoxidation of enone substrates, and indeed any Michael reaction of unsaturated carbonyl compounds catalyzed by 1a, structural information about the iminium intermediate of type

pathways. The quoted free energy differences are either taken directly from the BP86/def2-SVP results or represent best estimates derived from single-point B3LYP/def2-TZVP data with all aforementioned corrections included. Preliminary calculations revealed that the counteranion and hydrogen peroxide generally approach the iminium ion from the same side, allowing the anion to take up the proton liberated during this reaction step. We chose to use the iminium ion derived from 9-amino(9-deoxy)epicinchonidine 1d, which corresponds to 1a with an unsubstituted quinoline group, and enal 19e, together with TFA, as our model system (Figure 4).57 We considered four different pathways in the TS modeling: the quinoline group can adopt two different orientations (up and down) during the approach of the anion and hydrogen peroxide from the (β)-si-face or (β)-re-face, giving rise to TSs TS-1 and TS-3 (si) or TS-2 and TS-4 (re). For each of these pathways, a series of starting geometries were screened at the BP86/def2-SVP level to locate the lowest-energy TS for hydrogen peroxide attack.58 For the peroxide approach from the (β)-si-face, the orientation of the quinoline group has little influence on the energetics, since the conformer with this group pointing toward the anion (TS-1, Figure 4) is only slightly favored (by 0.8 kcal/ mol, BP86) over the alternative one (TS-3) due to C−H···F hydrogen-bonding interactions with TFA (2.43−2.58 Å). The lowest-energy TS for peroxide attack from the (β)-re-face (TS2) also has the quinoline group pointing toward the anion, with the alternative orientation (TS-4) being rather strongly disfavored (by 4.3 kcal/mol, BP86). Importantly, both TSs for the (β)-si-face attack are lower in energy than those for (β)re-face attack. The relevant free energy difference between TS-1 and TS-2 amounts to 1.9 kcal/mol (BP86) or 2.4 kcal/mol (best B3LYP-based estimate). Several common features can be noted when comparing the two lowest-lying TSs that lead to the major and minor enantiomer, respectively (TS-1 and TS-2). In both cases, the peroxide O−H bond is broken, and the C−O bond is formed in a concerted way, resulting in a direct proton transfer to the counteranion. The second O−H bond of the nonattacking peroxide oxygen is also coordinated to the counteranion. Another common feature is the formation of a pseudofivemembered ring N1−C8−C9−H21−H, as already seen in the N

dx.doi.org/10.1021/ja402058v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society

Article

epoxidations. In addition to providing direct access to optically active peroxide and epoxide products, the described methods allow for further product functionalization to generate such useful building blocks as β-hydroxyketones, 1,2-dioxolanes, and epoxyalcohols without any erosion of the enantiomeric excess. Detailed mechanistic investigations provide extensive support for covalent aminocatalysis and shed light on an unusual nondetrimental catalyst oxidation pathway. In addition, NMR and computational studies allowed for the rationalization of the origin of stereoselectivity induced by the ammonium salts of 1. Importantly, we have elucidated the role of the counteranion provided by the Brønsted acid cocatalyst, which translates remote chirality of the ammonium cation to the reaction center and functions as a base to activate the nucleophile. Given the excellent enantioselectivity and substrate scope, we anticipate utility of our methods in synthesis.

Figure 5. Proposed structure of monoprotonated iminium 27 derived from the N-oxidized catalyst 24.

A (Scheme 8) is required. In particular, with all other parameters being equal, the imine CN bond geometry is expected to determine the absolute facial selectivity of the Michael addition. Unfortunately, attempts to synthesize imine intermediates from enones for conformational studies did not meet with success. However, computational studies by Chen and colleagues on the epoxidation of 2-cyclohexen-1-one 9a by chiral cinchona alkaloid-derived amine salts58 and those of Zhu, Liu, and co-workers on the epoxidation of the same substrate by a chiral primary diamine salt60 both suggest iminium intermediates with the Z-configured CN bond as the catalytically relevant species leading to the epoxide product. These results are in agreement with our stereochemical model, which correctly predicts the experimentally observed absolute configuration for all enone substrate classes (cyclic, acyclic, and macrocyclic), as long as the iminium intermediates are Zconfigured (Figure 6).



ASSOCIATED CONTENT

S Supporting Information *

Experimental part: procedures, compound characterization, NMR spectra, and GC and HPLC traces for all new compounds. Computational part: methods, detailed numerical results for energies, optimized Cartesian coordinates, and further discussion of the computational results. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

[email protected] Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This study has been generously supported by the Deutsche Forschungsgemeinsschaft (DFG), the Max-Planck-Gesellschaft (MPG), and the Fonds der Chemischen Industrie (FCI). We thank our GC, HPLC, mass spectrometry, and NMR departments as well as Marianne Hannappel, Simone Marcus, Petra Philipps, Natascha Wippich for technical assistance.

Figure 6. Stereoselectivity Mnemonic for Epoxidation Reactions of Unsaturated Carbonyl Compounds Catalyzed by Amine 1a.



While we do not yet have experimental proof for this mechanism of stereoselectivity with the enone substrates, the models depicted in Figure 6 can serve as a useful guide for predicting enantioselectivity. Indeed, we expect this mnemonic to be applicable to any Michael addition of protic nucleophiles to unsaturated carbonyl compounds catalyzed by 1a and its closely related derivatives, given the remarkable consistency in the stereochemical outcome reported in the literature.

REFERENCES

(1) Sharpless, K. B. Angew. Chem., Int. Ed. 2002, 41, 2024. (2) (a) Wynberg, H.; Greijdanus, B. J. Chem. Soc. Chem. Commun. 1978, 427. (b) Juila, S.; Guixer, J.; Masana, J.; Rocas, J.; Colonna, S.; Annuziata, R.; Molinari, H. J. Chem. Soc. Perkin Trans. 1 1982, 1317. (c) Katsuki, T.; Sharpless, K. B. J. Am. Chem. Soc. 1980, 102, 5976. (d) Irie, R.; Noda, K.; Ito, Y.; Matsumoto, N.; Katsuki, T. Tetrahedron Lett. 1990, 31, 7345. (e) Zhang, W.; Loebach, J. L.; Wilson, S. R.; Jacobsen, E. N. J. Am. Chem. Soc. 1990, 112, 2801. (f) Enders, D.; Zhu, J.; Raabe, G. Angew. Chem., Int. Ed. Engl. 1996, 35, 1725. (g) Yang, D.; Yip, Y.-C.; Tang, M.-W.; Wong, M.-K.; Zheng, J.-H.; Cheung, K.-K. J. Am. Chem. Soc. 1996, 118, 491. (h) Bougauchi, M.; Watanabe, S.; Arai, T.; Sasai, H.; Shibasaki, M. J. Am. Chem. Soc. 1997, 119, 2329. (i) Wang, Z.-X.; Tu, Y.; Frohn, M.; Zhang, J.-R.; Shi, Y. J. Am. Chem. Soc. 1997, 119, 11224. (j) Watanabe, S.; Arai, T.; Sasai, H.; Bougauchi, M.; Shibasaki, M. J. Org. Chem. 1998, 63. (k) Corey, E. J.; Zhang, F.-Y. Org. Lett. 1999, 1, 1287. (l) Jacques, O.; Richards, S. J.; Jackson, R. F. W. Chem. Commun. 2001, 2712. (m) Li, C.; Pace, E. A.; Liang, M. C.; Lobkovsky, E.; Gilmore, T. D.; Porco, J. A., Jr. J. Am. Chem. Soc. 2001, 123, 11308. (n) Berkessel, A.; Guixa, M.; Schmidt, F.; Neudörfl, J. M.; Lex, J. Chem.Eur. J. 2007, 13, 4483. (o) Erkkilä, A.; Pihko, P. M.; Clarke, M.-R. Adv. Synth. Catal. 2007, 349, 802. (p) Peris, G.; Jakobsche, C. E.; Miller, S. J. J. Am. Chem. Soc. 2007, 129, 8710. (q) Zhang, W.; Yamamoto, H. J. Am. Chem. Soc. 2007, 129, 286. For



CONCLUSIONS We have demonstrated that the readily available cinchona alkaloid-derived primary amines 1 serve as powerful catalysts in the asymmetric epoxidation and peroxidation of various carbonyl compounds with hydrogen peroxide. The reactions have been shown to tolerate a wide scope of aliphatic acyclic α,β-unsaturated ketones, small, medium and large ring-sized cyclic α,β-unsaturated ketones, and α-branched α,β-unsaturated aldehydes, giving rise to an enantiomeric excess of generally above 95:5 er. Stereoconvergence with respect to the substrate double-bond geometry and significant catalyst control over preexisting stereogenic centers is a feature of most of the presented O

dx.doi.org/10.1021/ja402058v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society

Article

reviews, see: (r) Porter, M. J.; Skidmore, J. Chem. Commun. 2000, 1215. (s) Adam, W.; Saha-Möller, C. R.; Ganeshpure, P. A. Chem. Rev. 2001, 101, 3499. (t) Lane, B. S.; Burgess, K. Chem. Rev. 2003, 103, 2457. (u) Xia, Q.-H.; Ge, H.-Q.; Ye, C.-P.; Liu, Z.-M.; Su, K.-X. Chem. Rev. 2005, 105, 1603. (v) Lattanzi, A. Curr. Org. Synth. 2008, 5, 117. (3) Weitz, E.; Scheffer, A. Chem. Ber. 1921, 54, 2344. (4) (a) Bunton, C. A.; Minkoff, G. J. J. Chem. Soc. 1949, 665. (b) Christian, C. F.; Takeya, T.; Szymanski, M. J.; Singleton, D. A. J. Org. Chem. 2007, 72, 6183. (c) House, H. O.; Ro, R. S. J. Am. Chem. Soc. 1958, 80, 2428. (d) Kelly, D. R.; Caroff, E.; Flood, R. W.; Heal, W.; Roberts, S. M. Chem. Commun. 2004, 2016. (e) Kelly, D. R.; Roberts, S. M. Pept. Sci. 2006, 84, 74. (f) Meerwein, H. J. Prakt. Chem. 1926, 113, 9. (g) Weitz, E.; Scheffer, A. Chem. Ber. 1921, 54, 2344. (h) Wright, P.; Abbot, J. Int. J. Chem. Kinet. 1993, 25, 901. It was later shown that for metallated peroxides, the reaction sometimes proceeds through a concerted asynchronous manner. For a discussion, see: (i) Švenda, J.; Myers, A. G. Org. Lett. 2009, 11, 2437 However we are not aware of any such observation for the organocatalytic variant of the reaction. (5) (a) Marigo, M.; Franzen, J.; Poulsen, T. B.; Zhuang, W.; Jørgensen, K. A. J. Am. Chem. Soc. 2005, 127, 6964−6965. (b) Sunden, H.; Ibrahem, I.; Cordova, A. Tetrahedron Lett. 2006, 47, 99. (c) Lee, S.; MacMillan, D. W. C. Tetrahedron 2006, 62, 11413. (6) (a) Wang, X.; List, B. Angew. Chem., Int. Ed. 2008, 47, 1119. For reviews on ACDC, see: (b) Mahlau, M.; List, B. Angew. Chem., Int. Ed. 2013, 52, 518. (c) Phipps, R. J.; Hamilton, G. L.; Toste, F. D. Nat. Chem. 2012, 4, 603. (d) Mahlau, M.; List, B. Isr. J. Chem. 2012, 52, 630. (e) Ratjen, L.; Müller, S.; List, B. Nachr. Chem. 2010, 58, 640. (7) For a single example of secondary amine catalysis used in the epoxidation of α-branched acroleins, see: (a) Bondzic, B. P.; Urushima, T.; Ishikawa, H.; Hayashi, Y. Org. Lett. 2010, 12, 5434. For examples using noncovalent organocatalysis for Weitz−Scheffertype epoxidation, see: (b) Lattanzi, A. Org. Lett. 2005, 7, 2579. (c) Russo, A.; Galdi, G.; Croce, G.; Lattanzi, A. Chem.Eur. J. 2012, 18, 6152. (8) (a) Martin, N. J. A.; List, B. J. Am. Chem. Soc. 2006, 128, 13368. (b) Reisinger, C. M.; Wang, X.; List, B. Angew. Chem., Int. Ed. 2008, 47, 8112. (c) Wang, X.; Reisinger, C. M.; List, B. J. Am. Chem. Soc. 2008, 130, 6070. (d) Zhou, J.; Wakchaure, V.; Kraft, P.; List, B. Angew. Chem., Int. Ed. 2008, 47, 7656. (e) Lifchits, O.; Reisinger, C. M.; List, B. J. Am. Chem. Soc. 2010, 132, 10227. (f) Jiang, G.; List, B. Angew. Chem., Int. Ed. 2011, 50, 9471. (g) Lee, A.; Michrowska, A.; SulzerMosse, S.; List, B. Angew. Chem., Int. Ed. 2011, 50, 1707. (h) Lee, A.; List, B. Adv. Synth. Catal. 2012, 354, 1701. (9) (a) Ishihara, K.; Nakano, K. J. Am. Chem. Soc. 2005, 127, 10504. (b) Huang, H.; Jacobsen, E. N. J. Am. Chem. Soc. 2006, 128, 7170. (c) Lalonde, M. P.; Chen, Y.; Jacobsen, E. N. Angew. Chem., Int. Ed. 2006, 45, 6366−6370. (d) Kim, H.; Yen, C.; Preston, P.; Chin, J. Org. Lett. 2006, 8, 5239. (e) Ishihara, K.; Nakano, K. J. Am. Chem. Soc. 2007, 129, 8930. (f) Xie, J.-W.; Chen, W.; Li, R.; Zeng, M.; Du, W.; Yue, L.; Chen, Y.-C.; Wu, Y.; Zhu, J.; Deng, J.-G. Angew. Chem., Int. Ed. 2007, 46, 389. (10) (a) Peng, F.; Shao, Z. J. Mol. Catal. A: Chem. 2008, 285, 1. (b) Xu, L. W.; Luo, J.; Lu, Y. Chem. Commun. 2009, 1807. (c) Xu, L.W.; Lu, Y. Org. Biomol. Chem. 2008, 6, 2047. (11) (a) Chen, Y.-C. Synlett 2008, 13, 1919. (b) Jiang, L.; Chen, Y.-C. Catal. Sci. Technol. 2011, 1, 354. (c) Yeboah, E. M. O.; Yeboah, S. O.; Singh, G. S. Tetrahedron 2011, 67, 1725. (d) Li, J.-L.; Fu, N.; Zhang, L.; Zhou, P.; Luo, S.; Cheng, J.-P. Eur. J. Org. Chem. 2010, 35, 6840. (e) Lu, X.; Liu, Y.; Sun, B.; Cindric, B.; Deng, L. J. Am. Chem. Soc. 2008, 130, 8134. (f) Lu, Y.; Zheng, C.; Yang, Y.; Zhao, G.; Zou, G. Adv. Synth. Catal. 2011, 353, 3129. (g) Melchiorre, P. Angew. Chem., Int. Ed. 2012, 51, 9748. (12) For the original report describing preparation of 1a, see: Brunner, H.; Bügler, J.; Nuber, B. Tetrahedron: Asym. 1995, 6, 1699. (13) Organocatalytic methodologies for the peroxidation and epoxidation of aliphatic enones have apeared since; see refs 11e and 11f.

(14) (a) Bentley, P. A.; Bergeron, S.; Cappi, M. W.; Hibbs, D. E.; Hursthouse, M. B.; Nugent, T. C.; Pulido, R.; Roberts, S. M.; Wu, L. E. Chem. Commun. 1997, 739. (b) Daikai, K.; Hayano, T.; Kino, R.; Furuno, H.; Kagawa, T.; Inanaga, J. Chirality 2003, 15, 83. (c) Ooi, T.; Ohara, D.; Tamura, M.; Maruoka, K. J. Am. Chem. Soc. 2004, 126, 6844. (15) (a) Cativiela, C.; Figueras, F.; Fraile, J. M.; García, J. I.; Mayoral, J. A. Tetrahedron Lett. 1995, 36, 4125. (b) Fraile, J. M.; García, J. I.; Mayoral, J. A.; Sebti, S.; Tahirb, R. Green Chem. 2001, 3, 271. (c) Payne, G. J. Org. Chem. 1958, 23, 310. (d) Rieche, A.; Schmitz, E.; Gründemann, E. Chem. Ber. 1960, 93, 2443. (16) (a) Barnier, J.-P.; Morisson, V.; Blanco, L. Synth. Commun. 2001, 31, 349. (b) Kulinkovich, O. G.; Astashko, D. A.; Tyvorskii, V. I.; Ilyina, N. A. Synthesis 2001, 10, 1453. (17) See Supporting Information for further details. (18) Other hydroxy-group containing substrates were also tested. For details on these experiments, see the Supporting Information. (19) (a) Yang, J. W.; Hechavarria Fonseca, M. T.; Vignola, N.; List, B. Angew. Chem., Int. Ed. 2005, 44, 108. (b) Ouellet, S. G.; Tuttle, J. B.; MacMillan, D. W. C. J. Am. Chem. Soc. 2005, 127, 32. (20) For further support of the suggested mechanism from substrate isomerization and studies on β,β-disubstituted substrates, see the Supporting Information. (21) Dakin, H. D. Am. Chem. J. 1909, 42, 477. (22) (a) Bartoli, G.; Bosco, M.; Carlone, A.; Pesciaioli, F.; Sambri, L.; Melchiorre, P. Org. Lett. 2007, 9, 1403. (b) Xie, J.-W.; Huang, X.; Fan, L.-P.; Xu, D.-C.; Li, X.-S.; Su, H.; Wen, Y.-H. Adv. Synth. Catal. 2009, 351, 3077. (23) Fehr, C.; Galindo, J. Helv. Chim. Acta 2005, 88, 3128. (24) For the reduction of 3-hydroxy-1,2-dioxolanes with trivalent phosphorus compounds, see: (a) Baumstark, A. L.; Vasquez, P. C.; Chen, Y. Heteroat. Chem. 1993, 4, 175. (b) Morisson, V.; Barnier, J.-P.; Blanco, L. Tetrahedron Lett. 1999, 40, 4045. (25) Carlone, A.; Bartoli, G.; Bosco, M.; Pesciaioli, F.; Ricci, P.; Sambri, L.; Melchiorre, P. Eur. J. Org. Chem. 2007, 5492. (26) For the synthesis of aldol-like products 6 from epoxy ketones, see the Supporting Information. (27) (a) Dai, P.; Trullinger, T. K.; Liu, X.; Dussault, P. H. J. Org. Chem. 2006, 71, 2283. (b) Martyn, D. C.; Beletsky, G.; Cortese, J. F.; Tyndall, E.; Liu, H.; Fitzgerald, M. M.; O’Shea, T. J.; Liang, B.; Clardy, J. Bioorg. Med. Chem. Lett. 2009, 19, 5657. (c) Oh, D.-C.; Scott, J. J.; Currie, C. R.; Clardy, J. Org. Lett. 2009, 11, 633. (d) Scott, J. J.; Oh, D.-C.; Yuceer, M. C.; Klepzig, K. D.; Clardy, J.; Currie, C. R. Science 2008, 322, 63. (28) For synthetic approaches to racemic 1,2-dioxolanes, see: (a) Dussault, P. H.; Lee, I. Q.; Lee, H.-J.; Lee, R. J.; Niu, Q. J.; Schultz, J. A.; Zope, U. R. J. Org. Chem. 2000, 65, 8407. (b) Dussault, P. H.; Lee, R. J.; Schultz, J. A.; Suh, Y. S. Tetrahedron Lett. 2000, 41, 5457. (c) Dussault, P. H.; Liu, X. Org. Lett. 1999, 1, 1391. (d) Dussault, P. H.; Trullinger, T. K.; Cho-Shultz, S. Tetrahedron 2000, 56, 9213. (e) Dussault, P. H.; Zope, U. Tetrahedron Lett. 1995, 36, 3655. (f) McCullough, K. J.; Nojimab, M. Curr. Org. Synth. 2001, 5, 601. (g) Tokuyasu, T.; Ito, T.; Masuyama, A.; Nojima, M. Heterocycles 2000, 53, 1293. (29) Inanaga, J.; Kawanami, Y.; Yamaguchi, M. Bull. Chem. Soc. Jpn. 1986, 59, 1521. (30) For details on this transformation, see the Supporting Information. (31) Wynberg, H.; Marsman, B. J. Org. Chem. 1980, 45, 158. (32) For a discussion of this isomerization reaction, see the Supporting Information. (33) We also tested the more bulky t-Bu-substituted analog. For details, see the Supporting Information. (34) Zhou, J.; Wakchaure, V.; Kraft, P.; List, B. Angew. Chem., Int. Ed. 2008, 47, 7656. (35) (a) Bishara, A.; Rudi, A.; Goldberg, I.; Benayahu, Y.; Kashman, Y. Tetrahedron 2006, 62, 12092. (b) Shen, Y.-C.; Lin, J.-J.; Wu, Y.-R.; Chang, J.-Y.; Duh, C.-Y.; Lo, K. L. Tetrahedron Lett. 2006, 47, 6651. P

dx.doi.org/10.1021/ja402058v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society

Article

(36) Carey, F. A.; Sundberg, R. J. Advanced Organic Chemistry. Part A: Structure and Mechanisms: Structure and Mechanisms, 5 ed.; Springer: New York, 2007. (37) For the application of our epoxidation methodology in the synthesis of enantiomerically pure allylic alcohols, see: Jiang, H.; Holub, N.; Jørgensen, K. A. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 20630−20635. (38) (a) Pidathala, C.; Hoang, L.; Vignola, N.; List, B. Angew. Chem., Int. Ed. 2003, 42, 2785. (b) Ghobril, C.; Sabot, C.; Mioskowski, C.; Baati, R. Eur. J. Org. Chem. 2008, 4104. (39) (a) Hanessian, S.; Pham, V. Org. Lett. 2000, 2, 2975. (b) Mitchell, C. E. T.; Brenner, S. E.; Ley, S. V. Chem. Commun. 2005, 5346. (c) Perdicchia, D.; Jørgensen, K. A. J. Org. Chem. 2007, 72, 3565. (d) Wascholowski, V.; Knudsen, K. R.; Mitchell, C. E. T.; Ley, S. V. Chem.Eur. J. 2008, 14, 6155. (e) Li, P.; Wen, S.; Yu, F.; Liu, Q.; Li, W.; Wang, Y.; Liang, X.; Ye, J. Org. Lett. 2009, 11, 753. (f) Mase, N.; Fukasawa, M.; Kitagawa, N.; Shibagaki, F.; Noshiro, N.; Takabe, K. Synlett 2010, 2340. (g) Yoshida, M.; Hirama, K.; Narita, M.; Hara, S. Symmetry 2011, 3, 155. (40) For details on the optimization, see ref 8h. It is interesting to note that while Mosher’s acid proved to be the best acid additive, its stereoinformation had almost no influence on the selectivity of the reaction, with only 1% reduced ee in the mismatched case. (41) (a) Payne, G. B. J. Am. Chem. Soc. 1959, 81. (b) Payne, G. B. J. Org. Chem. 1961, 26, 205. (c) Zakavi, S.; Abasi, A.; Pourali, A. R.; Rayati, S. Bull. Korean Chem. Soc. 2008, 29, 866. (42) Organocatalytic methodologies for the epoxidation of branched acroleins have apeared since: see refs 6 and 11d. (43) (a) Galzerano, P.; Pesciaioli, F.; Mazzanti, A.; Bartoli, G.; Melchiorre, P. Angew. Chem., Int. Ed. 2009, 48, 7892. (b) Ishihara, K.; Nakano, K. J. Am. Chem. Soc. 2007, 129, 8930. (44) Colvin, E. W.; Robertson, A. D.; Wakharkar, S. J. Chem. Soc., Chem. Commun. 1983, 312. (45) In epoxidations of α-substituted acroleins 34, where the epoxide ring closure constitutes the enantiodetermining step, employing an achiral amine with a chiral Brønsted acid cocatalyst gives enantioenriched products. See also ref 6 for similar observations with 3,3-disubstituted enals.

(49) (a) Arai, S.; Tsuge, H.; Shioiri, T. Tetrahedron Lett. 1998, 39, 7563. (b) Lygo, B.; Gardiner, S. D.; McLeod, M. C.; To, D. C. M. Org. Biomol. Chem. 2007, 5, 2283. (c) Lygo, B.; Wainwright, P. G. Tetrahedron Lett. 1998, 39, 1599. (d) Lygo, B.; Wainwright, P. G. Tetrahedron 1999, 55, 6289. (e) Ooi, T.; Maruoka, K. Angew. Chem., Int. Ed. 2007, 46, 4222. (50) Oh, K.; Ryu, J. Tetrahedron Lett. 2008, 49, 1935. (51) Lu, N.; Mi, S.; Chen, D.; Zhang, G. Int. J. Quantum Chem. 2011, 111, 2874. (52) In contrast to the highly basic quinuclidine (pKa = 11.1 in water), tertiary amine N-oxides are only weakly basic (pKa around 4.5 in water). (53) (a) Cinchona Alkaloids in Synthesis and Catalysis; Song, C. E., Ed.; Wiley-VCH: Weinheim, 2009; (b) Silva, T. H. A.; Oliveira, A. B.; Santos, H. F. D.; Almeida, W. B. D. Struc. Chem. 2001, 12, 431. (c) Karle, J. M.; Karle, I. L.; Gerena, L.; Milhous, W. K. Antimicrob. Agents Chemother. 1992, 36, 1538. (d) Ferri, D.; Bürgi, T.; Baiker, A. J. Chem. Soc. Perkin Trans. 2 1999, 1305. (e) Dijkstra, G. D. H.; Kellogg, R. M.; Wynberg, H.; Svendsen, J. S.; Marko, I.; Sharpless, K. B. J. Am. Chem. Soc. 1989, 111, 8069. (f) Dijkstra, G. D. H.; Kellogg, R. M.; Wynberg, H. Recl. Trav. Chim. Pays-Bas 1989, 108, 195. (g) Bürgi, T.; Baiker, A. J. Am. Chem. Soc. 1998, 120, 12920. (54) For an example of specific ion pairing determined by heteronuclear NOESY techniques, see: (a) Olsen, R. A.; Borchardt, D.; Mink, L.; Agarwal, A.; Mueller, I. J.; Zaera, F. J. Am. Chem. Soc. 2006, 128, 15594. Hydrogen bonding interaction between the Brønsted acid and imine 41 also cannot be excluded: (b) Fleischmann, M.; Drettwan, D.; Sugiono, E.; Rueping, M.; Gschwind, R. M. Angew. Chem. 2011, 123, 6488. (55) Similar observations were made in iminium-catalyzed intramolecular aza-Michael reaction of enones employing catalyst 1a, where ethanol solvent also resulted in a lower and reversed enantioselectivity of the product compared to THF: Liu, J.-D.; Chen, Y.-C.; Zhang, G.B.; Li, Z.-Q.; Chen, P.; Du, J.-Y.; Tu, Y.-Q.; Fan, C.-A. Adv. Synth. Catal. 2011, 353, 2721. (56) (a) Díaz-Torres, R.; Alvarez, S. Dalton Trans. 2011, 40, 10742. (b) Kotlewska, A. J.; van Rantwijk, F.; Sheldon, R. A.; Arends, I. W. C. E. Green Chem. 2011, 13, 2154. (57) Catalyst 1d was found to give similar enantioselectivity as catalyst 1a but has less conformational flexibility, making it ideally suited for computational purposes. Specifically, when the epoxidation of (E)-31a (eq 5) was carried out using 1d under otherwise identical conditions, the results were close to identical. Epoxialdehyde 32a was obtained in 44% conversion, 73:27 dr and 92:8 er (38% yield, 75:25 dr and 93:7 er with 1a). (58) All calculations were carried out using TURBOMOLE. See Supporting Information for computational details. (59) Liu, X.-H.; Lin, L.-L.; Feng, X.-M. Acc. Chem. Res. 2011, 44, 574. (60) Lv, P.-L.; Zhu, R.-X.; Zhang, D.-J.; Duan, C.-G.; Liu, C.-B. J. Phys. Chem. A 2012, 116, 1251.

(46) An exception to this is the epoxidation of β,β-substituted enones 7/10 and α-substituted acroleins 34, where the initial Michael addition of hydrogen peroxide does not create a stereogenic center; it is the subsequent enamine-catalyzed ring closure that consitutes the enantiodetermining step of the reaction. (47) We also submitted racemic peroxyhemiketal 3a to the reaction conditions and obtained epoxide 4a with only a low enantiomeric excess (63:37 er) resulting from from a weak kinetic resolution at 44% conversion. These results are incompatible with a reversible mechanistic regime, which would be expected to furnish the epoxide 4a with > 98.5:1.5 er via dynamic kinetic resolution, i.e. the er value observed under actual reaction conditions (see Scheme 4 and Table 2).

(48) To this goal we compared 1a to a series of structurally distinct catalysts. For details, see the Supporting Information. Q

dx.doi.org/10.1021/ja402058v | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX